Heritage Financial (HFWA): Absorption Arithmetic on Puget Sound

Heritage Bank runs a commercial trade from Olympia north through the Puget Sound corridor, southwest into the Willamette Valley through the Eugene office, and east into Boise through the branch opened in the early part of the decade. The Seattle and Portland metros anchor both sides of the loan book on balance, and the deposit base follows most of the way. Population projections for those metros through the end of the decade exceed the national rate on every reading in the company's own materials, and household income there sits well above the national median. A lender tied to those counties owns a growth tide that does most of the rising without issuance help. Projected household income growth through the end of the decade runs double the national reading in the Seattle and Boise metros, the two fastest pools in the branch map. Unemployment across the branch metros sits at or below the national reading, which cushions the credit side of the same map.

The acquisition ledger maps the strategy forward as well as backward. Six whole-bank deals since the start of the last decade, each one bolted onto a Washington or Oregon branch web. Washington Banking arrived in mid-decade and roughly doubled the network, Puget Sound and Premier Commercial closed late in the same cycle, and Olympic adds roughly a billion and a half on top of a seven billion platform. The deal trail shows a decade spent compounding small and whole at once, and the pattern prices discipline over size. Every whole-bank add arrived from a neighboring geography the company already understood, which keeps integration inside a known regulatory and credit culture. The binding choice in the Kitsap integration keeps the acquired branch names in place, a trade of brand unity for customer retention in the towns that made the book.

Two structural features separate this franchise from the average regional of its size, and both show inside the funding stack rather than the loan stack. The company paid down borrowed funding during the first half while the median peer renewed it, which shows in the year-over-year decline in wholesale interest expense even after the advance drawn at quarter end. The fact that carries the shareholder return yield runs through a chart of deposits per branch, and the acquired base enters as the densest slice of the portfolio, and branch count rose from fifty to sixty six through the acquisition while deposits per branch declined only modestly.

The strategic frame resolves into two growth engines. Organic hiring in the big metros adds commercial books without branch capital, while the acquired base holds the suburbs and peninsulas between them. Seattle metro deposits ran near two and three quarter billion at midyear and Portland deposits near a billion, which leaves the two anchors covering over half the funding map. The Spokane branch opened in May, the Boise office compounds inside its second full rate cycle, and the combined branch foot of sixty six doors carries the merged deposit base into a map the strategy documents describe as unfinished consolidation. Both engines draw on the same pricing advantage the company built on relationship credit at a scale where lists of names matter more than national platforms, and each deal the company closes hands it a denser branch web to feed the next cycle.

The lending book splits into commercial real estate near half the total, a quarter in owner occupied collateral, and a commercial and industrial book just under a fifth. Restaurant and hotel paper inside that book ran criticized through the pandemic years and came through the cycle without visible loss severity. The rest of the collateral map splits between owner occupied industrial and flex space across the metro suburbs, which keeps single names small and appraisal coverage dense. Inside the non owner owned slice, industrial carries the biggest share, multifamily sits in the mid teens, and office runs near a tenth with risk ratings holding near the middle of the scale across every collateral type. The loan tape reads close to the corridor's industrial economy rather than to its speculative construction, and appraisal coverage on the commercial slices runs dense enough to hold valuations honest through a downturn.

Geography carries the distribution. Branches spread across four metro areas with density weighted toward the Seattle side, and the acquired Kitsap cluster enters the map at over two hundred thousand per branch on average. The arms of the loan map reach Eugene and Spokane, while deposit clusters sit along harbor towns that anchor smaller trade areas. Commercial lending in the Seattle metro ran near two and three quarter billion at midyear, which places the anchor market carrying near half the loan map alone. Liquidity in this book speaks through coverage: available funding sources cover estimated uninsured deposits by more than their full measure, with a wide margin to spare.

The entry story runs on people rather than product. Teams of commercial bankers moved into Eugene, Boise, and Spokane in consecutive years, each bringing books of relationships rather than an acquisition, and the same pattern earlier added specialty teams in builder lending and healthcare credit inside the home metros. The Boise branch opened in early decade, the Spokane branch opened this year, and both markets project faster growth than the coastal base. This model adds a production office wherever a hire warrants one and a full branch follows once the book supports it. New loan commitments rebounded across the second quarter after a winter trough, which leaves the production engines growing books again inside the merged footprint.

Funding flexibility covers the rest of the moat picture, though the scale understates how little it gets used. Borrowed funds run under a third of a billion against a seven billion deposit base, and the average deposit cost stays under one and a third on the whole stack. The bank carries a documented right to originate new advances at the Federal Home Loan Bank and the Federal Reserve, and it barely touches either during normal operation. Brokered capacity of roughly one and a fifth billion sits documented under internal policy, sized above the entire uninsured layer of the combined balance sheet. The moat reads as a density trade where early movers into inland metros carry relationship pricing a late buyer cannot match at the same cost, and where retired branch brands stay on the door to hold a customer base that never asked for a rebrand. Cost of total deposits lands inside the top decile of banks nationwide on the company's own comparative reading, which is the kind of funding evidence a rebrand would put at risk.

The second-quarter print ran at net income of seventeen and a half million with diluted earnings of forty two cents on a GAAP basis. Adjusted earnings of fifty seven cents per share excluded merger costs of about seven and a half million and a small securities loss, and the adjusted figure reached the best quarterly haul in company history. The margin story carries the franchise inside the quarter: net interest margin came in near four percent, up several basis points sequentially and nearly fifty basis points above the prior-year quarter. Deposit costs fell during the same window as acquired Kitsap accounts repriced lower, and the loan yield held above five and a half percent. First half net income ran almost forty percent above the prior-year window, a jump built on the merged balance sheet carrying a full two quarters at the better margin. Return on average tangible equity adjusted for merger noise ran near thirteen percent for the quarter, which extends a three-year average just above twelve.

Under the margin, the fixed-rate loan book carries the repricing story forward. Roughly half the loan book floats or reprices on schedule, and the fixed remainder carries maturities stacked beyond five years, with only about a tenth repricing inside a year. The nearest resets step from the mid five percent range toward the mid six percent range, a slower climb than the headline frame implies. The securities yield moved above three and a half percent after reinvesting maturities, adding a second engine to the same spread road. Certificates run near a seventh of deposits at a blended rate near three and a quarter, and every maturity rolling into the cash channel reprices the stack downward. Growth arrived by purchase in the first quarter and flattened in the second: the acquired book added roughly nine hundred million at the January close, then the whole company grew loans by a quarter of a percent through June. Line utilization on the commercial book sat in the mid fifties at the June date, which leaves room for funded growth without new intake, and construction commitments on the book run near three quarters of a billion against two thirds of a billion loans drawn.

The expense side tells the real cost of the deal. Merger costs of five and two tenths million in the first quarter and seven and a half in the second pushed the reported efficiency ratio into the mid seventies, while the adjusted ratio ran in the low sixties. Those dollar figures line up against full-time equivalents of nine hundred seventy six on a base of seven hundred forty a year earlier, and the headcount arithmetic sets the post-conversion efficiency target beyond the pre-deal trend line. The data processing line doubled quarter over quarter on the acquired account count, an artifact of running two core platforms through the bridge, and platform consolidation remains the single largest mechanical expense release left on the calendar. A record quarter with a flat book and a larger branch web leaves the expense release as the one piece of underwriting still pending, and the bridge quarter before conversion carries the heaviest duplicate load of the whole calendar.

The credit picture carries almost no narrative load for once in company history. Classified assets sit at the lowest reading in company records, nonaccrual loans run under a third of a percent of loans, and the provision line produced two consecutive negative quarters. The company carries an allowance at just above one percent of loans, sized on a portfolio whose charge-off average stayed near zero through the pandemic period, and that allowance runs near four hundred percent of the small nonaccrual balance underneath it. Credit quality works here as passive engine rather than cleanup narrative, and the cost of carrying that allowance drops in direct proportion. Charge offs against the acquired book were absorbed in purchase accounting at the close, which leaves the organic portfolio starting the post-merger period with an untouched allowance. Roughly one part in eight of the uninsured layer carried a full public pledge at the June date, which trims the funding beta on any deposit stress event.

Four events decide how the acquisition arithmetic lands on the stock, and each works through a stated mechanism rather than sentiment. The Olympic merger closed at the end of January in an all-stock exchange at a fixed ratio of forty five shares for each Olympic share. Because the ratio carried no collar, the consideration value moved with the market between the September signing and the closing date, and the acquired shareholders ended the trade roughly five percent richer than announcement week implied. Mechanically, the exchange added roughly seven million shares to the count, carried the obligation for a fixed dividend on seven million additional shares, and left the buyer's earnings denominator holding all of the acquired base at the higher closing price. Each side carried a reciprocal termination fee of seven million at signing, which priced the walk-away risk on both boards during the stock move between announcement and close.

The accounting standard changeover landed in January under early adoption of the new purchase accounting treatment for acquired loans. The bank split the acquired book into pools with and without identified credit deterioration, and the non-deteriorated pool carried its entire fair value discount into the accretion layer rather than into reserves. Most of the acquired book priced as purchased loans without identified deterioration, so the fair value discount flowed through accretion rather than a day-one reserve, and the small deteriorated tranche carried its own separate allowance at the close. That choice helped the margin print above four percent with the first month of the acquired book reflected, and the mechanism decays mathematically: accretion rides a shrinking discount base that amortizes down each quarter while the underlying loan yield carries the story back toward the ex-accretion level. The core deposit intangible on the acquired base amortizes over a ten-year life and feeds the same optical drag on reported expense every quarter until the combined platform earns past the bridge.

The dividend raise arrived in late July, the second increase in twelve months, and marked the first board action funded by the combined balance sheet. The mechanism matters more than the raise itself: the board matched the increase to combined earnings rather than to deal ramp assumptions, which leaves the combined dividend obligation carrying the acquired share count at roughly forty one million shares for the next four quarters. Borrowed funds entered the June quarter above one hundred sixty million after sitting near twenty million in the first quarter, and the movement covers inherited maturing brokered certificates and seasonal outflows in one stroke. The inherited brokered stack matured into a quarter when seasonal balances move, and the bank chose cash plus a short advance over renewal at current rates, which prices the funding bottom against a Wholesale alternative rather than a customer concession. The second-half deposit print reveals whether the combined base grows organically or the bank pays wholesale rates to stand still. The funding team sized a brokered renewal decline against a deposit base that carries over a quarter of itself in accounts paying nothing at all, which keeps the blended cost inside the low one and a quarter range even while wholesale doors stayed shut.

Three variables carry the thesis. Variable one, accretion decay: the acquired-loan discount amortizes toward zero and carries net interest income down with it, while the fixed-rate book repricing pushes the other way, and the net path sets the margin trajectory. Variable two, post-conversion expense release: the savings estimate requires the combined bank to shed duplicate platform cost in the second half, and the adjusted efficiency ratio inside the low sixties stands as the visible test. Variable three, acquired-deposit retention: the Kitsap base carries a below-average funding cost, and the balance that stays after the name plate carries the margin improvement.

The biggest single risk sits inside the CRE mix that defines this franchise. Owner occupied commercial real estate runs near a fifth of loans and non owner occupied near double that share, with office paper inside the bigger bucket at roughly six hundred eighty five million as of the June report. Vacancy surveys across the downtown Seattle and Bellevue submarkets put office availability well above pre-pandemic norms, and the acquired Kitsap book carries its own small office slice, with hotel and retail collateral holding the criticized share of the acquired map at levels the company tracked and reserved through the pandemic period. Reserves against that segment run near the same single digit percent as the general portfolio, which leaves little cushion inside the allowance if valuations compress. The portfolio concentration pattern mirrors the funding map, dense in metros with post-pandemic office recut risk.

The fixed-rate loan repricing cuts both ways. A falling rate path pulls income down through the same machinery that lifts it under a hold, since the floating half of the book resets with the short end and the flattening curve cuts reinvestment yields on the securities book at the same time. The fixed portion carries maturities stacked beyond five years at the lowest coupons on the book, and roughly a tenth of the total reprices inside a year at stepups toward the mid six percent range. The swap book is negligible at the June date, which leaves the margin path unhedged by design, a management choice that assumes repricing steadies the book through whatever the rate cycle brings.

Deposit behavior in the acquired Kitsap base carries behavioral risk that a rate-sensitive money market sweep can disturb, and conversion timing sits in the same window as the retention verification. Uninsured balances run near two fifths of the combined deposit total, and the liquidity stack covers that layer above its full line by a wide margin. If a sweep captures even a small tranche of acquired commercial balances, the post-close margin improvement depends on buying that funding back with borrowed money at a higher rate. A messy conversion creates exactly the customer service friction that drives balance migration, and a slip pushes duplicate data processing cost of roughly five million per quarter into the next year. One event carries two downstream damage channels, one on the expense line and one on the funding line.

The estimate band on the sell side runs from two thirty nine up to two eighty per share for the combined year, and the stock carries a market multiple near twelve times against the low end. Any shortfall in the post-conversion quarter lands directly on that multiple, and the rate path feeds the same variable from the repricing side. The stock sits near the top third of its own year range, which leaves little multiple cushion against a single weak print. The bear construction leans on the same chain: the conversion slips into the next year, the accretion falls faster than repricing adds, and a deposit tranche departs at the worst moment, which is why the stock multiple carries so little forgiveness into a single weak print.

The right lens for a mid-cycle bank holding acquired earnings is a double bar: price against tangible book value and price against merger-adjusted earnings power. Price to tangible book is the anchor because the merger costs distort exactly four quarters of earnings, while book value absorbs the deal at cost. The market cap to tangible book at spot sits near one and a half times after the September run, which places the stock in the upper band of the Pacific Northwest and Pacific coast regional cohort. History buys part of that premium: adjusted returns on tangible equity averaged near twelve percent across the prior three years, credit losses stayed near zero through a full rate cycle, and the deal added a lower cost funding base. Institutional registers hold most of the float, and six analysts attach published estimates to the combined company.

One structural fact of the merger window matters for the premium: the market paid a full multiple for the buyer's franchise during the earnings pause. Adjusted earnings per share of fifty seven cents in the merger quarter annualizes near two thirty on the combined share count, and the market cap divided by that run rate lands near twelve times annualized earnings, which brackets the low end of the sell side band for the combined year. About half the book carries index-linked pricing into any rate turn, which leaves the combined year estimate more sensitive to the rate path than a fixed-rate lender would carry. The reported figure compares at a wider number until the merger costs roll off, which makes the optical print look more expensive than the underlying case. That optical gap matters at a small float, because a single weak headline quarter moves the multiple faster than any strategic argument moves it back.

The buyback arithmetic puts a floor under the frame. The company spent ten million buying back shares during the second quarter at prices near one and a half times tangible book, a pace that annualizes near three and a half percent of the market cap. Combined with the raised dividend, the total shareholder return yield runs north of six percent while the balance sheet holds a pipeline of credit growth unfunded. The stock spent the last year ranging between the low twenty one level reached in November and an early thirty level set at the July high, which frames the current print near the top third of its own range. A franchise repurchasing stock above book defends that behavior with earned returns above twelve percent on tangible equity, and the operating history shows exactly that. The regular dividend has now risen every year of the recent span, with special payments layered on top in earlier cycles when capital ran over the well capitalized line.

The scenarios resolve against three cases anchored in the named variables, each measured from the September print near twenty eight and a half. Roughly two thirds of a billion in securities cash flows land inside the next three years, all reinvestable at yields above the portfolio run rate, which supports the base case earning power from the funding side. The bear case pairs a one times tangible book multiple with a tangible book around twenty and a half after two years of deal drag, which puts the shares near twenty, roughly thirty percent below the September print. The base case pairs the peer median of one point four times with tangible book compounding through the low twenties, which carries the shares toward thirty at the end of the run rate window, a few percent above spot. The bull case pairs a one point seven times multiple with a compounded tangible book near twenty two on the combination of margin strength and a second bolt-on, which carries the shares toward thirty eight, and the consolidation math on the company's own count shows dozens of regional banks still small enough to trade whole. The premium survives the framework honestly: the stock trades above the local band because the adjusted returns clear twelve percent and the funding base improved by acquisition, and the gap between the local band and the premium equals the difference between a documented twelve percent return franchise and a middle of the pack regional earning single digits.

Heritage Financial enters the descent off its own record quarter with the market holding a premium multiple against the combined book. The structural case rests on spread expansion through repricing, a funding base improved by acquisition, and a balance sheet with several hundred million in liquidity. The tactical case rests on the systems conversion in the current quarter and the expense release that carries the earnings improvement.

The judgment here treats the Olympic integration as a completed acquisition risk with an open expense settlement. The board funding the raised dividend from combined earnings implies confidence in the savings estimate, and the clean credit picture removes the usual provision noise that turns bank quarters into coin flips. The counterargument takes a simpler form: the market entry price already books the well run regional premium, so favorable execution adds little and execution failure subtracts plenty. What remains open is mechanism: the conversion actually shedding duplicate cost, the Kitsap base staying at its low funding cost, and the accretion decay being outrun by repricing. Those are three separate management execution tests rather than one deal bet, and they settle inside the next two quarterly reports.

The program of paying the dividend, executing buybacks, and holding capital at the well capitalized range constrains the upside arithmetic to a franchise that compounds. Nothing in the balance sheet suggests pressure toward the bear case near the visible horizon, and the downside materials hide inside the CRE bucket and the conversion calendar rather than anywhere in the clean credit book. The stock carries the premium, and the earnings line carries the burden through the second half.

The clean print tells a different story than the GAAP line. Purchase accounting pushed the acquired loans below par, and the discount streams back into income across the remaining life of the book. The margin prints near four percent with the accretion inside it and nearer three point nine percent excluding it, which keeps the underlying run rate visible rather than buried. Accretion decay runs against the fixed-rate repricing schedule through the same quarters, and the net of those two paths becomes the margin call that carries or breaks the premium. The accretion contribution ran near one and three quarter million in the latest quarter, sized small enough to decay inside a half decade on normal prepayment behavior. The conversion story carries the second half: merger costs consumed about twelve and a half million during the first half, and the systems changeover in the current quarter ends the duplicated platform spending that pushed the reported efficiency ratio into the mid seventies. Everything the premium requires flows through one settlement window, and the first post-conversion expense print settles the whole debate in either direction.